CN105069429B - A kind of flow of the people analytic statistics methods and system based on big data platform - Google Patents
A kind of flow of the people analytic statistics methods and system based on big data platform Download PDFInfo
- Publication number
- CN105069429B CN105069429B CN201510465923.9A CN201510465923A CN105069429B CN 105069429 B CN105069429 B CN 105069429B CN 201510465923 A CN201510465923 A CN 201510465923A CN 105069429 B CN105069429 B CN 105069429B
- Authority
- CN
- China
- Prior art keywords
- people
- target
- information
- data
- flow
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/53—Recognition of crowd images, e.g. recognition of crowd congestion
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of flow of the people analytic statistics methods and system based on big data platform, gather the video stream data of target scene and obtain video frame images, video frame images are carried out with format conversion and obtains target image and target image is labeled；The people information of target image sequence is analyzed according to target detection track algorithm, obtains the flow of the people data of target scene and the information data of people in target scene；By the information data transmission of people in the flow of the people data and target scene of target scene to big data platform；Analytic statistics is carried out to the information data of people in the flow of the people data and target scene of target scene based on big data platform and obtains people flow rate statistical result and the Information Statistics result of people.Analytic statistics is carried out to the information data of people in the flow of the people data and target scene of target scene based on big data platform, realizes people flow rate statistical and the displaying of the local scene of target scene and the global area of multiple target scenes.
Description
Technical field
The present invention relates to big data technical field, more particularly to a kind of flow of the people analytic statistics side based on big data platform
Method and system.
Background technology
The public places such as crossing, bank, market carry out people flow rate statistical, people flow rate statistical data can be used as build,
The important evidence of management, decision-making etc., for example, statistics and analysis by flow of the people data, we can more science close
Reason ground layout means of transportation, management bank's safety, decision-making market marketing mode etc..
Existing people flow rate statistical method mainly has three kinds, one is manually checking people by scene or monitor video
Number, the second is by the signal statistics flow of the people of mobile equipment, the third is counting flow of the people by Video Analysis Technology.It is existing
Most of people flow rate statistical method is all offline, it does not accomplish that the flow of the people of the real-time overall situation visualizes, and the global stream of people in real time
Amount visualization can provide authentic data for management and decision-making.
The content of the invention
Based on technical problem existing for background technology, the present invention proposes a kind of flow of the people analysis based on big data platform
Statistical method and system, realize the flow of the people system of the local scene of target scene and the global area of multiple target scenes
Meter and analysis.
A kind of flow of the people analytic statistics methods based on big data platform proposed by the present invention, including：
S1, the video stream data for gathering target scene simultaneously obtain video frame images, and format conversion is carried out to video frame images
Obtain target image and target image is labeled；
S2, according to target detection track algorithm analyze the people information of target image sequence, obtains target scene
Flow of the people data and target scene in people information data, the wherein information of people includes position of the people in target scene, big
Small, speed, direction, color；
S3, by the information data transmission of people in the flow of the people data and target scene of target scene to big data platform；
S4, based on big data platform divide the information data of people in the flow of the people data and target scene of target scene
Analysis statistics obtains people flow rate statistical result and the Information Statistics result of people.
Wherein, in S1, the video stream data of the collection target scene simultaneously obtains video frame images, specifically includes：It is logical
The video stream data of video acquisition module collection target scene is crossed, video frame images are captured from video stream data and obtain video
The image information of two field picture；
Preferably, the format conversion that carried out to video frame images obtains target image and target image is labeled,
Specifically include：Format conversion is carried out to video frame images according to format conversion algorithms and obtains rgb format target image, and marks mesh
The time tag and sequence label of logo image.
It is wherein, described that the people information of target image sequence is analyzed according to target detection track algorithm in S2,
Specifically include：According to target image sequence, moving object detection is carried out using Background difference, and use connected region track algorithm
Motion target tracking is carried out with reference to Mean Sift particle filter algorithms, target scene is obtained according to target detection tracking result
The information data of people in flow of the people data and target scene；
Preferably, the flow of the people data of target scene are obtained according to target detection tracking result, specifically included：Examined in target
Survey during tracking to moving target marker number, the history stream of people in target scene is obtained according to the numbering statistics of moving target
Aggregate data and current time flow of the people data；
Preferably, the information data of people in target scene is obtained according to target detection tracking result, specifically included：According to mesh
Mark detecting and tracking result obtains the positional information and size information of target in the target image, according to former and later two sequence target figures
As in target positional information calculation target motion velocity information and judge the direction of motion information of target, according to the position of target
The color histogram that confidence breath and size information extract target from target image obtains the colouring information of target.
Wherein, in S3, by the information data transmission of people in the flow of the people data and target scene of target scene to big number
According to platform, specifically include：Information data by socket communications people in the flow of the people data and target scene of target scene
It is transferred to big data platform.
Wherein, in S4, analytic statistics is carried out to the flow of the people data of target scene based on big data platform and obtains the stream of people
Statistical result is measured, is specifically included：Flow of the people data analysis and statistics are carried out based on big data platform, according to any in target area
The flow of the people data of one target scene carry out analysis with count to obtain the history stream of people total amount of the target scene, any one when
Between the flow of the people of section, the flow of the people comparing result of any number of periods, flow of the people time changing curve as a result, according to target area
The flow of the people data of multiple target scenes carry out analysis with counting to obtain the history stream of people total amount, any one of the target area in domain
The flow of the people comparing result of a period multiple target scenes, the flow of the people time changing curve result of multiple target scenes；It is excellent
Selection of land, report is generated and for the inquiry and displaying of data according to people flow rate statistical result；
Preferably, in S4, analytic statistics is carried out to the information data of people in target scene based on big data platform and is obtained
The Information Statistics of people are as a result, specifically include：Based on information analysis of the big data platform into pedestrian and statistics, according in target area
The information data of people carries out analysis and counts to obtain the information of historical stage people in the target scene in any one target scene
Statistical result, the Information Statistics result of any one period people, the information comparing result of any number of period people, according to mesh
The information data of people carries out analysis and counts to obtain historical stage people in the target area in multiple target scenes in mark region
The information data comparing result of people in Information Statistics result, any one period multiple target scenes；Preferably, according to people's
Information Statistics result generation report is used for the inquiry and displaying of data；
Preferably, big data platform stores data using the Hdfs of hadoop file system.
Wherein, S5 is further included after S4：Machine learning algorithm based on big data platform to people flow rate statistical result and
The Information Statistics result of people carries out flow of the people feature mining and the information characteristics of people excavate；
Preferably, the flow of the people feature mining includes being calculated according to people flow rate statistical result and the Information Statistics result of people
Flow of the people Density Distribution Q, the Q=Count/S of target scene, wherein Count is global area or regional area in target scene
Number, S be target scene in the area of global area or regional area；
Preferably, the information characteristics of the people excavate the size information and velocity information meter included according to people in target scene
Calculate the attribute of people, it is preferable that the age attribute of people is calculated by bayesian algorithm, Bayes's calculation formula is as follows：P (s | x)=p
(s)p(x|s)/(p(s)p(x|s)+p(b)p(x|b))；
Wherein, p (s | x) is judged as the probability of adult for people x, and p (s) is the probability being grown up, p occur under target scene
(b) it is to occur the probability of children under target scene, and p (x | s) according to the height with sample of being grown up and average movement velocity calculating phase
Like degree gained, and p (x | b) according to obtained by the height and average movement velocity with children's sample calculate similarity；Pass through given threshold
T, when p (s | x) >=T is then judged as being grown up, when p (s | x) ＜ T are then judged as children.
The invention also provides a kind of flow of the people analytic statistics system based on big data platform, including：
Video acquisition module, for gathering the video stream data of target scene；
Data processing module, for obtaining video frame images according to the video stream data of target scene, to video frame images
Format conversion is carried out to obtain target image and be labeled target image；
Image analysis module, for being divided according to target detection track algorithm the people information of target image sequence
Analysis, obtains the flow of the people data of target scene and the information data of people in target scene, and the wherein information of people includes people in target
Position, size in scene, speed, direction, color；
Big data console module, the Information Number for people in the flow of the people data and target scene of the target scene to transmission
People flow rate statistical result and the Information Statistics result of people are obtained according to analytic statistics is carried out.
Wherein, data processing module is specifically used for：Video frame images are captured from video stream data and obtain video frame figure
The image information of picture；
Preferably, data processing module is specifically used for：Format conversion is carried out to video frame images according to format conversion algorithms
Obtain rgb format target image, and the time tag and sequence label of label target image.
Wherein, image analysis module is specifically used for：According to target image sequence, moving target inspection is carried out using Background difference
Survey, and motion target tracking is carried out using connected region track algorithm combination Mean Sift particle filter algorithms, examined according to target
Survey tracking result and obtain the flow of the people data of target scene and the information data of people in target scene；
Preferably, image analysis module is specifically used for：To moving target marker number, root during target detection tracking
The history stream of people aggregate data and current time flow of the people data in target scene are obtained according to the numbering statistics of moving target；
Preferably, image analysis module is specifically used for：Target is obtained according to target detection tracking result in the target image
Positional information and size information, according in former and later two sequence target images the positional information calculation target of target movement speed
Degree information simultaneously judges the direction of motion information of target, and mesh is extracted from target image according to the positional information of target and size information
Target color histogram obtains the colouring information of target.
Wherein, image analysis module is communicated by socket modes with big data console module is used for the people of target scene
The information data transmission of people is to big data console module in data on flows and target scene；Preferably, big data platform uses
The Hdfs storage data of hadoop file system；
Preferably, big data console module is specifically used for：According to the flow of the people of any one target scene in target area
Data carry out analysis with counting to obtain the history stream of people total amount of the target scene, flow of the people of any one period, any more
Flow of the people comparing result, the flow of the people time changing curve of a period is as a result, according to multiple target scenes in target area
Flow of the people data carry out analysis and count to obtain the history stream of people total amount of the target area, any one period multiple target fields
The flow of the people comparing result of scape, the flow of the people time changing curve result of multiple target scenes；Preferably, according to people flow rate statistical
As a result report is generated and for the inquiry and displaying of data；
Preferably, big data console module is specifically used for：According to the letter of people in any one target scene in target area
Breath data carry out analysis and count to obtain the Information Statistics result of historical stage people in the target scene, any one period people
Information Statistics result, the information comparing result of any number of period people, according to people in multiple target scenes in target area
Information data carry out analysis with count to obtain the Information Statistics result of historical stage people in the target area, any one time
The information data comparing result of people in the multiple target scenes of section；Preferably, report is generated according to the Information Statistics result of people to be used for
The inquiry and displaying of data；
Preferably, big data console module is additionally operable to：Machine learning algorithm based on big data platform is to people flow rate statistical
As a result the information characteristics for flow of the people feature mining and people being carried out with the Information Statistics result of people excavate；
Preferably, big data console module is used to calculate target according to people flow rate statistical result and the Information Statistics result of people
Flow of the people Density Distribution Q, the Q=Count/S of scene, wherein Count is the people of global area or regional area in target scene
Number, S are the area of global area or regional area in target scene；
Preferably, big data console module is used to calculate people's according to the size information and velocity information of people in target scene
Attribute, it is preferable that the age attribute of people is calculated by bayesian algorithm, Bayes's calculation formula is as follows：P (s | x)=p (s) p (x
|s)/(p(s)p(x|s)+p(b)p(x|b))；
Wherein, p (s | x) is judged as the probability of adult for people x, and p (s) is the probability being grown up, p occur under target scene
(b) it is to occur the probability of children under target scene, and p (x | s) according to the height with sample of being grown up and average movement velocity calculating phase
Like degree gained, and p (x | b) according to obtained by the height and average movement velocity with children's sample calculate similarity；Pass through given threshold
T, when p (s | x) >=T is then judged as being grown up, when p (s | x) ＜ T are then judged as children.
In the present invention, by gathering the video stream data of target scene and obtaining video frame images, to video frame images into
Row format is converted to target image, and the people information of target image sequence is analyzed according to target detection track algorithm,
Obtain the flow of the people data of target scene and the information data of people in target scene, the people based on big data platform to target scene
The information data of people carries out analytic statistics and obtains people flow rate statistical result and the Information Statistics of people in data on flows and target scene
As a result；By technical scheme, based on big data platform to people in the flow of the people data and target scene of target scene
Information data carry out analytic statistics, realize the local scene of target scene and the global area of multiple target scenes
People flow rate statistical and displaying, and the information of people can be counted and collected；Machine learning algorithm based on big data platform
Flow of the people feature mining can be carried out to the Information Statistics result of people flow rate statistical result and people and the information characteristics of people excavate.
Brief description of the drawings
Fig. 1 is a kind of flow diagram of flow of the people analytic statistics methods based on big data platform proposed by the present invention.
Fig. 2 is a kind of catenation principle figure of flow of the people analytic statistics system based on big data platform proposed by the present invention.
Embodiment
With reference to Fig. 1, Fig. 1 is a kind of flow of the flow of the people analytic statistics methods based on big data platform proposed by the present invention
Schematic diagram.
As shown in Figure 1, a kind of flow of the people analytic statistics methods based on big data platform proposed by the present invention, including：
S1, the video stream data for gathering target scene simultaneously obtain video frame images, and format conversion is carried out to video frame images
Obtain target image and target image is labeled；
S2, according to target detection track algorithm analyze the people information of target image sequence, obtains target scene
Flow of the people data and target scene in people information data, the wherein information of people includes position of the people in target scene, big
Small, speed, direction, color；
S3, by the information data transmission of people in the flow of the people data and target scene of target scene to big data platform；
S4, based on big data platform divide the information data of people in the flow of the people data and target scene of target scene
Analysis statistics obtains people flow rate statistical result and the Information Statistics result of people.
In a particular embodiment, in S1, target scene is gathered by video acquisition module (such as web camera)
Video stream data.Video frame images are captured from the video stream data of collection, video flowing are split into video frame, and obtain video
The image information of two field picture, wherein image information include picture format, image size, frame number information, temporal information etc..According to lattice
Formula transfer algorithm carries out format conversion to video frame images, so that the target image of rgb format is obtained, and label target image
Time tag and sequence label.
Realized when carrying out image format conversion according to corresponding format conversion algorithms, specifically, according to conversion
Front and rear picture format requires to select corresponding format conversion algorithm, and in the prior art, image format conversion is highly developed
Solution, this will not be detailed here.
For example, when YUV image is converted into RGB image, its format conversion algorithms is as follows：
R=Y+1.402 (Cr-128),
G=Y-0.34414 (Cb-128) -0.71414 (Cr-128),
B=Y+1.772 (Cb-128).
In a particular embodiment, in S2, the target image sequence based on multiple target images carries out the inspection of moving target
Tracking is surveyed, moving object detection is carried out using Background difference, and using the filter of connected region track algorithm combination Mean Sift particles
Ripple algorithm carries out motion target tracking.Background difference is a kind of most common method, the work of Background difference in current motion segmentation
Make principle and realize that process may refer to Liyuan Li, Weimin Huang, Irene Y.H.Gu, Qi Tian,
" Foreground object detection from videos containing complex background ",
Proceeding MULTIMEDIA'03Proceedings of the eleventh ACM international
conference on Multimedia Pages 2-10ACM New York,NY,2003.Target following is using " even
Logical area tracking algorithm " combination " Mean Sift particle filter algorithms " carries out crash analysis, sees to solve the tenacious tracking of target
Problem and collision separation problem.
During target detection tracking, moving target is numbered, can be counted according to the numbering of moving target
Obtain the history stream of people aggregate data and current time flow of the people data in target scene.The result of target detection tracking is exactly mesh
Position and the size of target image are marked on, the fortune of target can be calculated according to the position of target in former and later two sequence target images
Dynamic speed simultaneously judges the direction of motion of target, extracts the color histogram of target from target image according to the position of target and size
Figure, so as to obtain position of the people in target scene, size, speed, direction, color, that is, obtains the information of people in target image.
In a particular embodiment, in S3, by the information data of people in the flow of the people data and target scene of target scene
Big data platform is transferred to by socket communication modes, data message form is defined as follows in transmission：
Flow of the people data message defines：
Time | video camera | accumulative flow of the people | current flow of the people |
The information data message definition of people：
Time | video camera | numbering | color | position | size | direction | speed |
In a particular embodiment, big data platform stores data using the Hdfs of hadoop file system, in distributed ring
Data can be read with concurrent fashion under border and carry out data processing.
In a particular embodiment, in S4, analysis system is carried out to the flow of the people data of target scene based on big data platform
Meter obtains people flow rate statistical as a result, specifically including：Carried out according to the flow of the people data of any one target scene in target area
Analysis and statistics, can obtain the history stream of people total amount of the target scene, the flow of the people of any one period, it is any number of when
Between the flow of the people comparing result of section, flow of the people time changing curve result etc.；According to the people of multiple target scenes in target area
Data on flows is analyzed and counted, and can obtain history stream of people total amount, any one period multiple mesh of the target area
Mark flow of the people comparing result, the flow of the people time changing curve result of multiple target scenes of scene；According to people flow rate statistical knot
Fruit generates report and is used for inquiry and displaying of data etc.；
In a particular embodiment, in S4, the information data of people in target scene is analyzed based on big data platform
Statistics obtains the Information Statistics of people as a result, specifically including：According to the Information Number of people in any one target scene in target area
According to being analyzed and being counted, Information Statistics result, any one period of historical stage people in the target scene can be obtained
The Information Statistics result of people, the information comparing result etc. of any number of period people；According to multiple target scenes in target area
The information data of middle people is analyzed and counted, and can be obtained the Information Statistics result of historical stage people in the target area, be appointed
Information data comparing result of people etc. in one period multiple target scene of meaning；Report is generated according to the Information Statistics result of people
Inquiry and displaying for data.
In a particular embodiment, S5 is further included after S4：Machine learning algorithm based on big data platform is to flow of the people
Statistical result and the Information Statistics result of people carry out flow of the people feature mining and the information characteristics of people excavate.Existing big data is put down
Many existing machine learning algorithms are integrated with platform, flow of the people can be excavated using machine learning algorithm according to user demand
The characteristic information of statistical result and the Information Statistics result of people.For example, by the position of people, size, speed, side in target scene
To, color as input, feature mining is carried out to the attribute of flow of the people Density Distribution, people by corresponding machine learning algorithm.
For example, flow of the people feature mining includes the Information Statistics result (position of people according to people flow rate statistical result and people
Information) calculate target scene flow of the people Density Distribution Q；Q1 and Q2 is distributed as whole locals and local area in target scene
Flow of the people Density Distribution data, wherein Q1=Count1/S1, Count1 are the number of global area in target scene, and S1 is mesh
The area of global area in scene is marked, alternatively, Q2=Count2/S2, wherein Count2 are the people of regional area in target scene
Number, S1 are the area of regional area in target scene；
Include calculating people's according to the size information and velocity information of people in target scene for example, the information characteristics of people excavate
Attribute, the age attribute of people is calculated by bayesian algorithm, and Bayes's calculation formula is as follows:
P (s | x)=p (s) p (x | s)/(p (s) p (x | s)+p (b) p (x | b))；
Wherein, p (s | x) is judged as the probability of adult for people x, and p (s) is that the probability for occurring being grown up under target scene (can
With as obtained by artificial statistical sample), p (b) (can be by artificial statistical sample institute to occur the probability of children under target scene
), p (x | s) according to and adult's sample height with average movement velocity calculate similarity obtained by, p (x | b) according to and children's sample
This height is calculated obtained by similarity with average movement velocity；By given threshold T, when p (s | x) >=T is then judged as being grown up, when
P (s | x) ＜ T are then judged as children.
With reference to Fig. 2, Fig. 2 is a kind of flow of the flow of the people analytic statistics system based on big data platform proposed by the present invention
Schematic diagram.
As shown in Fig. 2, a kind of flow of the people analytic statistics system based on big data platform proposed by the present invention, including：
Video acquisition module, for gathering the video stream data of target scene；
Data processing module, communicates with video acquisition module, for obtaining video according to the video stream data of target scene
Two field picture, carries out video frame images format conversion and obtains target image and target image is labeled；
Image analysis module, communicates with data processing module, for according to target detection track algorithm to target image sequence
The people information of row is analyzed, and obtains the flow of the people data of target scene and the information data of people in target scene, wherein people
Information include position of the people in target scene, size, speed, direction, color；
Big data console module, communicates with image analysis module, for the target scene to transmission flow of the people data and
The information data of people carries out analytic statistics and obtains people flow rate statistical result and the Information Statistics result of people in target scene.
Wherein, data processing module is specifically used for：Video frame images are captured from video stream data and obtain video frame figure
The image information of picture, image information include picture format, image size, frame number information, temporal information etc.；Calculated according to format conversion
Method carries out format conversion to video frame images and obtains rgb format target image, and the time tag and sequence of label target image
Label.
Wherein, image analysis module is specifically used for：According to target image sequence, moving target inspection is carried out using Background difference
Survey, and motion target tracking is carried out using connected region track algorithm combination Mean Sift particle filter algorithms, examined according to target
Survey tracking result and obtain the flow of the people data of target scene and the information data of people in target scene；Process is tracked in target detection
In to moving target marker number, according to the numbering statistics of moving target obtain history stream of people aggregate data in target scene and
Current time flow of the people data；Obtain position and the size of target in the target image according to target detection tracking result, according to
The position of target calculates the movement velocity of target and judges the direction of motion of target in former and later two sequence target images, according to mesh
Target position and size extract the color histogram of target from target image, thus obtain position of the people in target scene,
Size, speed, direction, color.
Image analysis module is communicated by socket modes with big data console module to be used for the flow of the people of target scene
The information data transmission of people is to big data console module in data and target scene；Big data platform uses hadoop file system
Hdfs storage data；
Big data console module is specifically used for：Carried out according to the flow of the people data of any one target scene in target area
Analysis is with counting the history stream of people total amount for obtaining the target scene, the flow of the people of any one period, any number of periods
Flow of the people comparing result, flow of the people time changing curve is as a result, flow of the people number according to multiple target scenes in target area
According to carry out analysis with count to obtain the history stream of people total amount of the target area, any one period multiple target scenes the stream of people
Measure comparing result, the flow of the people time changing curve result of multiple target scenes；Report is generated simultaneously according to people flow rate statistical result
Inquiry and displaying for data；
Big data console module is specifically used for：According to the information data of people in any one target scene in target area into
Row analysis obtains the Information Statistics result of historical stage people in the target scene with statistics, the information of any one period people is united
Result, the information comparing result of any number of period people are counted, according to the Information Number of people in multiple target scenes in target area
Analyzed according to carrying out and count to obtain the Information Statistics result of historical stage people in the target area, any one period multiple mesh
Mark the information data comparing result of people in scene；Inquiry and exhibition of the report for data are generated according to the Information Statistics result of people
Show；
Big data console module is additionally operable to：Machine learning algorithm based on big data platform is to people flow rate statistical result and people
Information Statistics result carry out the information characteristics of flow of the people feature mining and people and excavate；
Big data console module is used to calculate target scene according to people flow rate statistical result and the Information Statistics result of people
Flow of the people Density Distribution Q；Q1 and Q2 is distributed as the flow of the people Density Distribution data of whole locals and local area in target scene,
Q1=Count1/S1, wherein Count1 are the number of global area in target scene, and S1 is the face of global area in target scene
Product, alternatively, Q2=Count2/S2, wherein Count2 are the number of regional area in target scene, S1 is local in target scene
The area in region；
Big data console module is used for the attribute that people is calculated according to the size information and velocity information of people in target scene, leads to
The age attribute that bayesian algorithm calculates people is crossed, Bayes's calculation formula is as follows：
P (s | x)=p (s) p (x | s)/(p (s) p (x | s)+p (b) p (x | b))；
Wherein, p (s | x) is judged as the probability of adult for people x, and p (s) is that the probability for occurring being grown up under target scene (can
With as obtained by artificial statistical sample), p (b) (can be by artificial statistical sample institute to occur the probability of children under target scene
), p (x | s) according to and adult's sample height with average movement velocity calculate similarity obtained by, p (x | b) according to and children's sample
This height is calculated obtained by similarity with average movement velocity；By given threshold T, when p (s | x) >=T is then judged as being grown up, when
P (s | x) ＜ T are then judged as children.
By technical scheme, based on big data platform in the flow of the people data and target scene of target scene
The information data of people carries out analytic statistics, realizes the local scene of target scene and the global area of multiple target scenes
People flow rate statistical and displaying, and the information of people can be counted and collected；Machine learning based on big data platform is calculated
Method can carry out the Information Statistics result of people flow rate statistical result and people flow of the people feature mining and the information characteristics of people excavate.
The foregoing is only a preferred embodiment of the present invention, but protection scope of the present invention be not limited thereto,
Any one skilled in the art the invention discloses technical scope in, technique according to the invention scheme and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (8)
- A kind of 1. flow of the people analytic statistics methods based on big data platform, it is characterised in that including：S1, the video stream data for gathering target scene simultaneously obtain video frame images, and carrying out format conversion to video frame images obtains Target image is simultaneously labeled target image；S2, according to target detection track algorithm analyze the people information of target image sequence, obtains the people of target scene The information data of people in data on flows and target scene, the wherein information of people include position of the people in target scene, size, speed Degree, direction, color；S3, by the information data transmission of people in the flow of the people data and target scene of target scene to big data platform；S4, based on big data platform carry out analysis system to the information data of people in the flow of the people data and target scene of target scene Meter obtains people flow rate statistical result and the Information Statistics result of people；It is described that the people information of target image sequence is analyzed according to target detection track algorithm in S2, specifically include： According to target image sequence, moving object detection is carried out using Background difference, and using connected region track algorithm combination Mean Sift particle filter algorithms carry out motion target tracking, and the flow of the people data of target scene are obtained according to target detection tracking result With the information data of people in target scene；The flow of the people data of target scene are obtained according to target detection tracking result, are specifically included：Process is tracked in target detection In to moving target marker number, according to the numbering statistics of moving target obtain history stream of people aggregate data in target scene and Current time flow of the people data；The information data of people in target scene is obtained according to target detection tracking result, is specifically included：Tracked according to target detection As a result the positional information and size information of target are obtained in the target image, according to target in former and later two sequence target images The motion velocity information of positional information calculation target and the direction of motion information for judging target, according to the positional information of target and greatly The color histogram that small information extracts target from target image obtains the colouring information of target.
- 2. the flow of the people analytic statistics methods according to claim 1 based on big data platform, it is characterised in that in S1 In, the video stream data of the collection target scene simultaneously obtains video frame images, specifically includes：Gathered by video acquisition module The video stream data of target scene, video frame images are captured from video stream data and obtain the image information of video frame images；The format conversion that carried out to video frame images obtains target image and target image is labeled, and specifically includes：Root Format conversion is carried out to video frame images according to format conversion algorithms and obtains rgb format target image, and label target image when Between label and sequence label.
- 3. the flow of the people analytic statistics methods according to claim 1 or 2 based on big data platform, it is characterised in that In S3, by the information data transmission of people in the flow of the people data and target scene of target scene to big data platform, specifically include： By socket communications the information data transmission of people in the flow of the people data and target scene of target scene to big data platform.
- 4. the flow of the people analytic statistics methods according to claim 1 or 2 based on big data platform, it is characterised in that In S4, based on big data platform the flow of the people data of target scene are carried out with analytic statistics and obtains people flow rate statistical as a result, specific Including：Flow of the people data analysis and statistics are carried out based on big data platform, according to any one target scene in target area Flow of the people data carry out analysis with count to obtain the history stream of people total amount of the target scene, the flow of the people of any one period, Flow of the people comparing result, the flow of the people time changing curve of any number of periods is as a result, according to multiple targets in target area It is multiple with counting to obtain the history stream of people total amount of the target area, any one period that the flow of the people data of scene carry out analysis The flow of the people comparing result of target scene, the flow of the people time changing curve result of multiple target scenes；According to people flow rate statistical As a result report is generated and for the inquiry and displaying of data；In S4, analytic statistics is carried out to the information data of people in target scene based on big data platform and obtains the Information Statistics of people As a result, specifically include：Based on information analysis of the big data platform into pedestrian and statistics, according to any one target in target area The information data of people carries out analysis with counting to obtain the Information Statistics result of historical stage people in the target scene, any in scene The Information Statistics result of one period people, the information comparing result of any number of period people, according to multiple in target area In target scene the information data of people carry out analysis with count to obtain the Information Statistics result of historical stage people in the target area, The information data comparing result of people in any one period multiple target scenes；Report is generated according to the Information Statistics result of people Inquiry and displaying for data；Big data platform stores data using the Hdfs of hadoop file system.
- 5. the flow of the people analytic statistics methods according to claim 1 or 2 based on big data platform, it is characterised in that S5 is further included after S4：Information Statistics result of the machine learning algorithm based on big data platform to people flow rate statistical result and people The information characteristics for carrying out flow of the people feature mining and people excavate；The flow of the people feature mining includes calculating target scene according to people flow rate statistical result and the Information Statistics result of people Flow of the people Density Distribution Q, Q=Count/S, wherein Count is the number of global area or regional area in target scene, and S is mesh Mark the area of global area or regional area in scene；The information characteristics of the people excavate the attribute for including that people is calculated according to the size information and velocity information of people in target scene, The age attribute of people is calculated by bayesian algorithm, Bayes's calculation formula is as follows：p（s|x）=p（s）p（x|s）/(p（s）p（x| s）+p（b）p（x|b））；Wherein, p（s|x）The probability of adult, p are judged as people x（s）To occur the probability being grown up, p under target scene（b）For Occur the probability of children, p under target scene（x|s）Similarity institute is calculated with average movement velocity according to the height with sample of being grown up , p（x|b）According to obtained by the height with children's sample and average movement velocity calculate similarity；By given threshold T, work as p（s |x）>=T is then judged as being grown up, and works as p（s|x）＜ T are then judged as children.
- A kind of 6. flow of the people analytic statistics system based on big data platform, it is characterised in that including：Video acquisition module, for gathering the video stream data of target scene；Data processing module, for obtaining video frame images according to the video stream data of target scene, carries out video frame images Format conversion obtains target image and target image is labeled；Image analysis module, for being analyzed according to target detection track algorithm the people information of target image sequence, obtains The information data of people into the flow of the people data and target scene of target scene, the wherein information of people include people in target scene Position, size, speed, direction, color；Big data console module, for people in the flow of the people data and target scene of the target scene to transmission information data into Row analytic statistics obtains people flow rate statistical result and the Information Statistics result of people；Image analysis module is specifically used for：According to target image sequence, moving object detection is carried out using Background difference, and use Connected region track algorithm combination Mean Sift particle filter algorithms carry out motion target tracking, are tracked and tied according to target detection Fruit obtains the flow of the people data of target scene and the information data of people in target scene；Image analysis module is specifically used for：To moving target marker number during target detection tracking, according to moving target Numbering statistics obtain the history stream of people aggregate data and current time flow of the people data in target scene；Image analysis module is specifically used for：According to target detection tracking result obtain in the target image target positional information and Size information, according to the motion velocity information of the positional information calculation target of target and judgement in former and later two sequence target images The direction of motion information of target, the color histogram of target is extracted according to the positional information of target and size information from target image Figure obtains the colouring information of target.
- 7. the flow of the people analytic statistics system according to claim 6 based on big data platform, it is characterised in that at data Reason module is specifically used for：Video frame images are captured from video stream data and obtain the image information of video frame images；Data processing module is specifically used for：Format conversion is carried out to video frame images according to format conversion algorithms and obtains rgb format Target image, and the time tag and sequence label of label target image.
- 8. the flow of the people analytic statistics system based on big data platform according to claim 6 or 7, it is characterised in that figure It is used for as analysis module is communicated by socket modes with big data console module by the flow of the people data and target of target scene The information data transmission of people is to big data console module in scene；Big data platform is stored using the Hdfs of hadoop file system Data；Big data console module is specifically used for：Analyzed according to the flow of the people data of any one target scene in target area With counting the history stream of people total amount for obtaining the target scene, the flow of the people of any one period, the people of any number of periods Current capacity contrast's result, flow of the people time changing curve as a result, according to the flow of the people data of multiple target scenes in target area into Row analysis obtains history stream of people total amount, the flow of the people pair of any one period multiple target scenes of the target area with statistics Than result, the flow of the people time changing curve result of multiple target scenes；Report is generated according to people flow rate statistical result and is used for The inquiry and displaying of data；Big data console module is specifically used for：Divided according to the information data of people in any one target scene in target area Analysis obtains Information Statistics result, the Information Statistics knot of any one period people of historical stage people in the target scene with statistics Fruit, the information comparing result of any number of period people, according to the information data of people in multiple target scenes in target area into Row analysis obtains Information Statistics result, any one period multiple target fields of historical stage people in the target area with statistics The information data comparing result of people in scape；Inquiry and displaying of the report for data are generated according to the Information Statistics result of people；Big data console module is additionally operable to：Letter of the machine learning algorithm based on big data platform to people flow rate statistical result and people Cease statistical result and carry out flow of the people feature mining and the information characteristics excavation of people；Big data console module is used for the stream of people that target scene is calculated according to people flow rate statistical result and the Information Statistics result of people Metric density is distributed Q, Q=Count/S, and wherein Count is the number of global area or regional area in target scene, and S is target field The area of global area or regional area in scape；Big data console module is used for the attribute that people is calculated according to the size information and velocity information of people in target scene, passes through shellfish This algorithm of leaf calculates the age attribute of people, and Bayes's calculation formula is as follows： p（s|x）=p（s）p（x|s）/(p（s）p（x|s）+p （b）p（x|b））；Wherein, p（s|x）The probability of adult, p are judged as people x（s）To occur the probability being grown up, p under target scene（b）For Occur the probability of children, p under target scene（x|s）Similarity institute is calculated with average movement velocity according to the height with sample of being grown up , p（x|b）According to obtained by the height with children's sample and average movement velocity calculate similarity；By given threshold T, work as p（s |x）>=T is then judged as being grown up, and works as p（s|x）＜ T are then judged as children.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510465923.9A CN105069429B (en) | 2015-07-29 | 2015-07-29 | A kind of flow of the people analytic statistics methods and system based on big data platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510465923.9A CN105069429B (en) | 2015-07-29 | 2015-07-29 | A kind of flow of the people analytic statistics methods and system based on big data platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105069429A CN105069429A (en) | 2015-11-18 |
| CN105069429B true CN105069429B (en) | 2018-05-15 |
Family
ID=54498791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510465923.9A Expired - Fee Related CN105069429B (en) | 2015-07-29 | 2015-07-29 | A kind of flow of the people analytic statistics methods and system based on big data platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105069429B (en) |
Families Citing this family (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105763853A (en) * | 2016-04-14 | 2016-07-13 | 北京中电万联科技股份有限公司 | Emergency early warning method for stampede accident in public area |
| CN106897743B (en) * | 2017-02-22 | 2020-05-05 | 广州市勤思网络科技有限公司 | Mobile attendance anti-cheating big data detection method based on Bayesian model |
| JP6918523B2 (en) * | 2017-03-06 | 2021-08-11 | キヤノン株式会社 | A program that causes a computer to execute an information processing system, an information processing device, an information processing method, and an information processing method. |
| CN108182403A (en) * | 2017-12-28 | 2018-06-19 | 河南辉煌城轨科技有限公司 | Subway train passenger flow statistical method based on image |
| CN110503284B (en) * | 2018-05-18 | 2022-08-05 | 杭州海康威视数字技术股份有限公司 | Statistical method and device based on queuing data |
| CN109101941A (en) * | 2018-08-26 | 2018-12-28 | 俞绍富 | Video monitoring management platform and its method |
| CN110532989B (en) * | 2019-09-04 | 2022-10-14 | 哈尔滨工业大学 | Automatic detection method for offshore targets |
| CN110738105A (en) * | 2019-09-05 | 2020-01-31 | 哈尔滨工业大学（深圳） | method, device, system and storage medium for calculating urban street cell pedestrian flow based on deep learning |
| CN111339873B (en) * | 2020-02-18 | 2021-04-20 | 南京甄视智能科技有限公司 | Passenger flow statistical method and device, storage medium and computing equipment |
| CN111383455A (en) * | 2020-03-11 | 2020-07-07 | 上海眼控科技股份有限公司 | Traffic intersection object flow statistical method, device, computer equipment and medium |
| CN111629335A (en) * | 2020-05-29 | 2020-09-04 | 四川亨通网智科技有限公司 | Method and system for realizing real-time people flow thermodynamic diagram of scenic spot based on big data |
| CN112417555B (en) * | 2020-11-18 | 2024-05-31 | 君泰天创工程咨询有限公司 | System and method for judging building design optimization scheme |
| CN112434101B (en) * | 2020-11-23 | 2021-06-25 | 北京航空航天大学 | System for carrying out people flow migration analysis by using shared trip big data |
| CN112995589A (en) * | 2020-12-28 | 2021-06-18 | 上海天跃科技股份有限公司 | Big data analysis system based on pedestrian flow statistical data |
| CN112801377B (en) * | 2021-01-29 | 2023-08-22 | 腾讯大地通途（北京）科技有限公司 | Object estimation method, device, equipment and storage medium |
| CN112885096A (en) * | 2021-02-05 | 2021-06-01 | 同济大学 | Bridge floor traffic flow full-view-field sensing system and method depending on bridge arch ribs |
| CN113163110B (en) * | 2021-03-05 | 2022-04-08 | 北京宙心科技有限公司 | People stream density analysis system and analysis method |
| CN113645180A (en) * | 2021-06-04 | 2021-11-12 | 复旦大学附属肿瘤医院 | People flow statistical analysis system and method thereof |
| CN113792688A (en) * | 2021-09-18 | 2021-12-14 | 北京市商汤科技开发有限公司 | Business state analysis method and device, electronic equipment and storage medium |
| CN114283386B (en) * | 2022-01-28 | 2024-06-21 | 浙江传媒学院 | Real-time monitoring system for analyzing and adapting to dense scene people stream based on big data |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902819A (en) * | 2012-10-30 | 2013-01-30 | 浙江宇视科技有限公司 | Intelligent video analysis method and device |
| CN103839308A (en) * | 2012-11-26 | 2014-06-04 | 中兴通讯股份有限公司 | Population obtaining method, device and system |
-
2015
- 2015-07-29 CN CN201510465923.9A patent/CN105069429B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102902819A (en) * | 2012-10-30 | 2013-01-30 | 浙江宇视科技有限公司 | Intelligent video analysis method and device |
| CN103839308A (en) * | 2012-11-26 | 2014-06-04 | 中兴通讯股份有限公司 | Population obtaining method, device and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105069429A (en) | 2015-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105069429B (en) | A kind of flow of the people analytic statistics methods and system based on big data platform | |
| CN101639354B (en) | Method and apparatus for object tracking | |
| CN104361327B (en) | A kind of pedestrian detection method and system | |
| CN108009473A (en) | Based on goal behavior attribute video structural processing method, system and storage device | |
| CN108053427A (en) | A kind of modified multi-object tracking method, system and device based on KCF and Kalman | |
| CN106203513B (en) | A kind of statistical method based on pedestrian's head and shoulder multi-target detection and tracking | |
| CN103473554B (en) | Artificial abortion's statistical system and method | |
| CN108052859A (en) | A kind of anomaly detection method, system and device based on cluster Optical-flow Feature | |
| Zhang et al. | Segmentation and tracking multiple objects under occlusion from multiview video | |
| CN106570449B (en) | A kind of flow of the people defined based on region and popularity detection method and detection system | |
| CN110084165A (en) | The intelligent recognition and method for early warning of anomalous event under the open scene of power domain based on edge calculations | |
| Ferryman et al. | Performance evaluation of crowd image analysis using the PETS2009 dataset | |
| CN102243765A (en) | Multi-camera-based multi-objective positioning tracking method and system | |
| US11055894B1 (en) | Conversion of object-related traffic sensor information at roadways and intersections for virtual dynamic digital representation of objects | |
| CN106156714A (en) | The Human bodys' response method merged based on skeletal joint feature and surface character | |
| CN107067755A (en) | A kind of method for calibrating traffic monitoring camera automatically based on computer vision | |
| CN103729620B (en) | A kind of multi-view pedestrian detection method based on multi-view Bayesian network | |
| CN105427345A (en) | Three-dimensional people stream movement analysis method based on camera projection matrix | |
| CN107480653A (en) | passenger flow volume detection method based on computer vision | |
| Jiang et al. | Data fusion-based multi-object tracking for unconstrained visual sensor networks | |
| CN114648557A (en) | Multi-target cooperative tracking method based on high-altitude visual angle and ground visual angle | |
| CN116597122A (en) | Data labeling method, device, electronic equipment and storage medium | |
| LU500512B1 (en) | Crowd distribution form detection method based on unmanned aerial vehicle and artificial intelligence | |
| CN113920585A (en) | Behavior recognition method and device, equipment and storage medium | |
| CN103530601A (en) | Monitoring blind area crowd state deduction method based on Bayesian network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180515 |